Overall Project Progress

[MisterBitcoin]

Originally Posted by sweety439

The count of the remain k's is not the same as the count of the remain k's in http://www.noprimeleftbehind.net/crus/tab/CRUS_tab.htm.

The remain k´s from CRUS_tab are the sum from all started bases/ranges.
KEP´s remain status is including the unstarted bases/ranges.

[KEP]

Originally Posted by MisterBitcoin

The remain k´s from CRUS_tab are the sum from all started bases/ranges.
KEP´s remain status is including the unstarted bases/ranges.

Correct

[masser]

Originally Posted by gd_barnes

We already have detailed stats for bases proven and bases with 1, 2, and 3 k's remaining:
http://www.noprimeleftbehind.net/cru...crus-stats.htm
http://www.noprimeleftbehind.net/crus/tab/CRUS_tab.htm

I too think that the total number of k's remaining on the project is not very relevant. Although I have kept a few files from testing for S280, I don't plan to ever keep track of either R280 or S280 on the pages. There are a few other huge conjectures that I don't plan to track.

I feel like KEP's statistic just encourages people to start huge bases. That is not a direction that I would like to see the project take. I would ask that people not start huge bases unless they plan to start them as part of their own side project.

+1. I like the stats pages Gary posted above. The only improvement might be if we added a row for "unsolvable" bases, but then we have problems defining what unsolvable means. Consider R6. I suspect the probability of finding a prime for that sequence is less than 1 now. Consider R3. As Gary mentioned, the final primes for that base won't be computable. R3 and R6 are very different cases, but both might be considered unsolvable.

[MisterBitcoin]

Originally Posted by masser

+1. I like the stats pages Gary posted above. The only improvement might be if we added a row for "unsolvable" bases, but then we have problems defining what unsolvable means. Consider R6. I suspect the probability of finding a prime for that sequence is less than 1 now. Consider R3. As Gary mentioned, the final primes for that base won't be computable. R3 and R6 are very different cases, but both might be considered unsolvable.

Well, I think both bases are solveable, but it´s a matter of time (and resources). It will take many years to do it, but it´s possible.
If BOINC would only take R3 we would see a massive progress on it. :)

[VBCurtis]

Originally Posted by MisterBitcoin

Well, I think both bases are solveable, but it´s a matter of time (and resources). It will take many years to do it, but it´s possible.
If BOINC would only take R3 we would see a massive progress on it. :)

Huh? BOINC has been working on R3 for over a year now, and we have seen massive progress. What are you talking about?

[MisterBitcoin]

Originally Posted by VBCurtis

Huh? BOINC has been working on R3 for over a year now, and we have seen massive progress. What are you talking about?

If there would be more persons for sieving on R3, BOINC would do more tests. The only person who sieves on R3 (sofar I saw it) are you.

[VBCurtis]

Originally Posted by MisterBitcoin

If there would be more persons for sieving on R3, BOINC would do more tests. The only person who sieves on R3 (sofar I saw it) are you.

Also false (well, OK, I am the only siever...). Reb nearly always has a sieve file from me in reserve from R3 that is not pushed to BOINC; there are simply more interesting things to do with those resources than to fill the channels with all R3.

I have the resources to keep BOINC running 50-100% faster on R3 than it does, but R3 just isn't a priority. We're making nice progress with the current setup, as are the multitude of other bases that Reb runs on BOINC.

[rebirther]

Originally Posted by VBCurtis

Also false (well, OK, I am the only siever...). Reb nearly always has a sieve file from me in reserve from R3 that is not pushed to BOINC; there are simply more interesting things to do with those resources than to fill the channels with all R3.

I have the resources to keep BOINC running 50-100% faster on R3 than it does, but R3 just isn't a priority. We're making nice progress with the current setup, as are the multitude of other bases that Reb runs on BOINC.

I will start some ranges for R3 soon. Due the lack of time in the past months I had no time to manage things.

[gd_barnes]

Originally Posted by MisterBitcoin

That´s an good idea. I´m willing to take care about S3/S7/S15, if you let me know what must be done.
Managing 1042 (x2) Bases is way to much for 1 person (even for two).

Good luck with that! It is far more effort than you will ever realize to get it completely right, even for one such huge base. You'll be dealing with huge primes and results files, lots of different people with different motivations, managing reservations, answering 100s of questions, dealing with many software changes that you have to make sure are right before they can be used, removing k's that have been primed, sorting primes and results, removing duplicated primes, and publicly posting and maintaining the status of the base, the k's remaining, and the reservations. I cannot begin to demonstrate the amount of effort involved.

I would definitely like CRUS to maintain the management of R3 and S3. If someone can demonstrate that they understand what is involved in managing a huge base and can demonstrate that they will stick with it for 5 years or longer and will not lose interest, I will gladly turn it over to them. I'm fine with turning over all of bases 7 and 15 and any other base with CK > 1G. Keep in mind that some of the huge bases are 10s to 100s of times larger than R3 and S3.

I am out of town now for another 10 days and don't really have time to explain in detail what needs to be done to manage a huge base. This is all that I have time to post. KEP has attempted the task. KEP can you talk with MisterBitcoin about possibly managing a huge base?

[gd_barnes]

Originally Posted by KEP

Something isn't quite right. On ALL odd bases ALL odd k's has been removed. However I cannot without testing all k's for trivial, GFN and MOB, remove any further k's.

It would be manageable, to remove the k's but it would mean about 700000+ spreadsheet tables would have to be created and that would require about 2100000+ minutes of work. So it could be managed, but it would make the stats much harder to do

Of course all odd k's on odd bases have been removed. They have a trivial factor of 2. Keep in mind that the starting bases script does not write odd k's to the trivial file for odd bases. It assumes that the user knows that such k's don't need to be tested.

Doing what you stated is a pointless task. If you want to track something meaningful for us, do this:
Look at all bases with a conjectured k less than 1M. Run all such unstarted bases through the starting bases script to get a close estimate of the number of k's that have to actually be primed for unstarted bases. Add those k's together with the k's remaining on all started bases and post those figures. That would be something that the project could reasonably shoot for.

[gd_barnes]

Originally Posted by MisterBitcoin

Well, I think both bases are solveable, but it´s a matter of time (and resources). It will take many years to do it, but it´s possible.
If BOINC would only take R3 we would see a massive progress on it. :)

I feel a little Silverman in me right now. This is ridiculous. It could not be done using all of the resources in the entire universe until the heat death of the known universe. Have you read this?:

http://www.mersenneforum.org/showpos...27&postcount=6

I don't have time to explain to you why this is the case if you cannot understand it. All that you have to do is some research into the exponentional increase in the size of the final primes on bases as well as the exponentional increase into the size of the testing time.

I'll tell you what...Why don't you reserve some really big tests on a base that has a chance to be proven? You seem to think that others can just find these large primes easily. How about reserving R31 that is currently tested to n=1M? Run that up to n=2M using all of the resources that you have and get back with me on solving bases. For that matter enlist 10 of your friends to help you search that base. See how long you all last with it.

I guess I get a little tired of demonstrating to people the size of the efforts needed to prove even smaller-conjectured bases.